An environment of large-scale modern networks is quite complex. In a process of network transmission, various abnormalities may occur. An OpenFlow (OpenFlow) network supports a controller to send, periodically or according to an instruction of an administrator, a query request to an OpenFlow switch to acquire related information, to detect, according to the acquired related information, whether an abnormality occurs in the OpenFlow network, and to process the abnormality when the abnormality occurs.
However, in the prior art, an abnormality that occurs in an OpenFlow network and is caused by an unexpected event, for example, network congestion caused by instantaneous excessive traffic, cannot be notified in time, so that the abnormality in the OpenFlow network cannot be processed in time and efficiency in processing the abnormality in the OpenFlow network is relatively low.